Looking for breakthrough ideas for innovation challenges? Try Patsnap Eureka!

Depth map and pose optimization method and system

An optimization method and depth map technology, applied in the field of computer vision, can solve problems such as unstable optimization process and poor perception of large targets, and achieve complete and accurate prediction results, clear and accurate prediction results, and reduce interference

Pending Publication Date: 2022-06-07
ADVANCED INST OF INFORMATION TECH AIIT PEKING UNIV +1
View PDF0 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0004] What the present invention aims to solve is that in the existing methods, the optimization process is unstable and the optimizer has poor perception of large targets.

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • Depth map and pose optimization method and system
  • Depth map and pose optimization method and system
  • Depth map and pose optimization method and system

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0040] as Figure 1 Shown, a depth map and pose optimization method and system based on grouped dual enhancement modules and multi-level perception optimizers, comprising the following steps.

[0041] Step 1, Data Preparation, divided into two steps:

[0042] 1.1, the preparation includes RGB image data, real depth map data and real camera posture data sequence, such as KITTI unmanned driving data set, which contains the car camera and a variety of sensors acquired image data, true depth map data and real camera posture data sequence.

[0043] 1.2, The target image It, the adjacent image of It (i from 1 to 2), the true pose Pi of It relative to Iri, and the true depth map of It are extracted from the preliminary data of 1.1. Preferably, the size of the image is preprocessed to be 320 pixels high and 960 pixels wide.

[0044] Step 2, image feature extraction and contextual information extraction. Utilize the Feature Extraction Module and context extraction module to It and Iri in St...

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

PUM

No PUM Login to View More

Abstract

The invention discloses a depth map and pose optimization method and system. The method comprises the following steps: preparing a data sequence; inputting the image data into an extractor, and respectively extracting image features and context information; obtaining an initial depth map, a pose and photometric measurement cost according to the image features; inputting the photometric measurement cost into a grouping double-enhancement module to obtain enhancement cost; the enhancement cost, the initial depth map and pose, and the context information are inputted into a multi-level perception optimizer, and an accurate depth map and pose are obtained after iterative optimization; and calculating an error according to a prediction result and a true value, and iteratively optimizing the whole deep neural network until the error converges.

Description

Technical field [0001] The present invention belongs to the field of computer vision, a depth map and pose optimization method and system based on the grouped double enhancement module and multi-level perception optimizer. Background [0002] Motion Recovery Structure (SfM) is designed to accurately calculate the depth map of a set of camera images and their relative poses to each other, thus modeling the three-dimensional structure of the corresponding scene. Traditional SfM optimization schemes generally rely on beam adjustment (BA) solutions. However, traditional BA methods require good image key points and feature matching between images, resulting in performance that is severely limited by lighting conditions and structural textures. [0003] In order to solve these problems, in recent years, some methods and systems have emerged to apply deep learning to solve SfM optimization problems. Early methods directly used neural networks to regress the input image to predict the ou...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

Application Information

Patent Timeline
no application Login to View More
Patent Type & Authority Applications(China)
IPC IPC(8): G06T7/73G06T3/40G06N3/04G06N3/08
CPCG06T7/73G06T3/4038G06N3/08G06T2207/10028G06T2207/20081G06T2207/20084G06T2200/32G06N3/045
Inventor 吴沛熹
Owner ADVANCED INST OF INFORMATION TECH AIIT PEKING UNIV
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Patsnap Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Patsnap Eureka Blog
Learn More
PatSnap group products

Browse by: Latest US Patents, China's latest patents, Technical Efficacy Thesaurus, Application Domain, Technology Topic, Popular Technical Reports.

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap|About US| Contact US: help@patsnap.com